Suppressing the cross terms of the Wigner Distribution with an adaptive frequency smoothing window

To extend the application of the fractional Fourier transform (FrFT) as well as to present an efficient method for analyzing texture images, a FrFT-based texture classification algorithm is proposed in this paper. 1-D local FrFTs along different orientations are designed to describe texture characteristics, while the minimum distance classifier with the χ2-static distance is adopted for classification. The superiority of the proposed method is examined and demonstrated through three groups of texture images.